Paresis of the upper limb after stroke results in limited functional movement ability for many patients. Recent interventions such as Constraint Induced Movement Therapy (CIT) have demonstrated strong effects on the recovery of functional movement for chronic stroke patients in the clinic. Assessing the carryover of clinical exercise therapy outcomes to the home and community is currently limited, however, to patient interview measures such as the Motor Activity Log. Objective remote measurement of functional movement performance could assist clinicians in evaluating the outcomes of CIT and other interventions for movement recovery after stroke. The PIs propose to collect ambulatory accelerometry (ACC) and electromyographic (EMG) data from healthy elderly and stroke subjects in order to quantify the functional amount of use (AOU) and quality of movement (QOM) in their paretic upper limbs. From the theoretical perspective that a small number of component submovements comprise all upper limb activities of daily living (ADL), the PIs hypothesize that with the ambulatory ACC/EMG system they will be able to identify when and for how long an ADL is performed by recognition of the presence of these submovements, or Functional Motor Tasks (FMT). Using previous studies that have characterized the quality of upper limb movement by patterns of kinematic coordination and muscle coactivation, they further hypothesize that analysis of the ACC/EMG data will reveal impairment level QOM features of FMT as well. They expect that this AOU and QOM data will correlate with clinical measures of upper limb impairment and function. Laboratory data will be collected from 5 healthy elderly and 20 post-stroke subjects performing FMT that are representative of UE movements necessary for the performance of ADL. Artificial neural networks will be trained to recognize the UE FMT and distinguish them from upper limb movements not involved with ADL (such as cyclical arm movement accompanying ambulation). Paretic upper limb QOM will be determined by identifying patterns of coordination such as EMG measures of muscle coactivation and kinematic features from ACC data. Further validation of the responsiveness of this system will be performed by recording ACC/EMG data of subjects performing a random sequence of ADLs in a non-laboratory setting. Subjects will be assessed using standard clinical scales of upper limb motor impairment and function such as the FugI-Meyer, the Wolf Motor Function Test, and the Motor Activity Log. The PIs anticipate that the technology for remote and unobtrusive ACC/EMG data collection will be available when they have completed this ambulatory study. Techniques developed in the pilot for monitoring paretic upper limb AOU and QOM will form the basis of a larger study. They will remotely monitor functional upper limb use outside the clinic in the homes and in the community of patients undergoing CIT.